A commonsense, supported analysis on the KOS & Nate Silver v. Research 2000 controversy from Richard Charnin (TruthIsAll)…
Regarding the Research 2000 Pile-on
(Also see the latest, Research 2000: A Closer Look at Volatility)
Richard Charnin (TruthIsAll)
Markos Moulitsas of Daily Kos and Nate Silver of fivethirtyeight.com have each taken aim at Research 2000, which did the pre-election polling for Daily Kos.
Markos cites the Research 2000 poll that determined only 27% of Republicans supported gays in the military. He found this to be way too low. His evidence? Gallup and ABC polls had over 60% GOP support. Well, because they are MSM, that makes them right and R2000 wrong?
He gives Republicans too much credit. If 90% of Democrats support gays in the military as do 30% of Republicans and 60% of independents, then we would have about 61% supporting in total. In any case, why does Markos go after R2000, which did the 2008 pre-election presidential polling for him?
Nate Silver wrote that the R2000 data “feels way too clean for me”. Better clean than dirty, Nate. Would he rather see the kind of volatility that CNN and Gallup had in their tracking polls? For some reason, Wolf Blitzer always appeared perplexed whenever Gore jumped in the 2000 poll.
Nate points to a chart depicting age breakdowns in the Democratic vote share for the last 20 contests surveyed by R2K and PPP. He writes: ”The age breakdowns in Research 2000’s numbers are almost always close to “perfect” — in 20 out of 20 cases, for instance, the Democrat gets a lower vote share from among 30-44 year olds than among 18-29 year olds. PPP’s data, on the other hand, is *much* messier — which is what I think we should expect when comparing small subsamples, particularly subsamples of lots of different races that are subject to different demographic patterns”.
Of course, 18-29 year olds consistently vote more Democratic than the 30-44 group. Is that news? Is 20 out of 20 cases reasonable? Let’s compute the probability that the Democratic share of the 18-29 age group would exceed their share of the 30-44 group in all 20 elections.
In 2008, Obama had 66% of the 18-29 segment and 52% of the 30-44 group. In 2004, Kerry had 56% and 48%, respectively. So in the last two elections, the Democrats had an average 61% share of the 18-29 age group and 50% of the 30-44 group. Given these shares, the probability is virtually 100% that all 20 elections would show that the 18-29 age group was the best one for the Democrats. Assuming a 56% Dem share of the younger group, the probability is 98% that the Dem 18-29 share would exceed the 30-44 share in all 20 elections.
Nate also finds it strange that in the last 30 races, Democrats did better among women in all cases. Once again, I ask, is that news? I will spare you the probability analysis. He also questions the lack of volatility in R2000. It’s not “noisy” enough for his taste.
“Likewise, take a look at their Presidential tracking numbers from 2008 (http://www.dailykos.com/dailypoll/2008/11/4). They published their daily results in addition to their three-day rolling average … and the daily results were remarkably consistent from day to day. At no point, for instance, in the two months that they published daily results did Obama’s vote share fluctuate by more than a net of 2 points from day to day (to reiterate, this is for the daily results (n=~360) and not the rolling average). That just seems extremely unlikely — there should be more noise than that.
You want noise? OK, Nate, let’s take a look at the national pre-election polls from Sept. 10 and compare Obama’s 3-poll moving average to the 3-day R2000 tracking poll. The average change in the R2Kpoll was 0.67% compared to 0.99% in the pre-election poll moving average. The moving average standard deviation was 1.58% for R2K compared to 1.88% for the national polls. The absolute standard deviation of the R2K percentage change in vote share was 0.53% compared to 0.58% for all polls Finally, the average Obama R2000 share was 50.27%. It was 49.65% in all polls.
The R2K poll shares are rounded to the nearest percent, so small changes are not reflected in the table. For instance, assume the actual R2K tracking poll results on successive days were 49.7% and 50.2%. These would both be shown as 50% by R2K and the actual 0.5% change would show as 0%. The effect is to lower the overall R2K volatility (standard deviation) to 1.58% as shown above. Therefore, the True R2K volatility must be closer to the ALL polls 1.88% figure.
Nate and Markos have each denigrated exit polls. In fact, they won’t even talk about Election Fraud. Yet they fabricate faux outrage about an independent polling firm that is not the MSM. Nate was recently hired by the NY Times. Unfortunately, the Grey Lady (“All the News that’s Fit to Print”) does not report Election Fraud. In 2004-2005, Markos locked out posters who sought to present statistical and anecdotal evidence pointing to election fraud. But the Kossacks rebelled. Those “conspiracy nuts” are now allowed to post.
Obama’s recorded share was 52.9%, a 9.5 million margin. But the True Vote model indicates he had 57.5%, a 22 million margin. Want proof that fraud cost Obama 13 million? Consider this: The National Exit Poll required that there be 12 million more returning Bush voters than Kerry voters. That is not only implausible; it is mathematically impossible. There were likely 10 million more returning Kerry voters than Bush voters.
Why don’t you write about it in the Times, Nate? Or mention it the next time you get on MSNBC? On the other hand, you better not – if you want to keep that job.